Epigenetic effects associated with salmonid supplementation and domestication

Several studies have demonstrated lower fitness of salmonids born and reared in a hatchery setting compared to those born in nature, yet broad-scale genome-wide genetic differences between hatchery-origin and natural-origin fish have remained largely undetected. Recent research efforts have focused on using epigenetic tools to explore the role of heritable changes outside of genetic variation in response to hatchery rearing. We synthesized the results from salmonid studies that have directly compared methylation differences between hatchery-origin and natural-origin fish. Overall, the majority of studies found substantial differences in methylation patterns and overlap in functional genomic regions between hatchery-origin and natural-origin fish which have been replicated in parallel across geographical locations. Epigenetic differences were consistently found in the sperm of hatchery-origin versus natural-origin fish along with evidence for maternal effects, providing a potential source of multigenerational transmission. While there were clear epigenetic differences in gametic lines between hatchery-origin and natural-origin fish, only a limited number explored the potential mechanisms explaining these differences. We outline opportunities for epigenetics to inform salmonid breeding and rearing practices and to mitigate for fitness differences between hatchery-origin and natural-origin fish. We then provide possible explanations and avenues of future epigenetics research in salmonid supplementation programs, including: 1) further exploration of the factors in early development shaping epigenetic differences, 2) understanding the functional genomic changes that are occurring in response to epigenetic changes, 3) elucidating the relationship between epigenetics, phenotypic variation, and fitness, and 4) determining heritability of epigenetic marks along with persistence of marks across generations.


Introduction
Salmonid species spanning the West Coast of the United States and upward toward southern British Columbia as well as those distributed along the East Coast of the United States have experienced extensive and ongoing declines throughout their native range (Parrish et al. 1998;Noakes et al. 2000;Gustafson et al. 2007). Numerous efforts have been enacted to Abstract Several studies have demonstrated lower fitness of salmonids born and reared in a hatchery setting compared to those born in nature, yet broad-scale genome-wide genetic differences between hatcheryorigin and natural-origin fish have remained largely undetected. Recent research efforts have focused on using epigenetic tools to explore the role of heritable changes outside of genetic variation in response to hatchery rearing. We synthesized the results from salmonid studies that have directly compared methylation differences between hatchery-origin and natural-origin fish. Overall, the majority of studies found substantial differences in methylation patterns and overlap in functional genomic regions between hatchery-origin and natural-origin fish which have been replicated in parallel across geographical locations. Epigenetic differences were consistently found in the sperm of hatchery-origin versus natural-origin fish along with evidence for maternal effects, providing a potential source of multigenerational transmission. While there were clear epigenetic differences in restore salmon runs, each meeting a unique series of challenges (Lichatowich 1999;Ruckelshaus et al. 2002;Allen 2003). One such effort has been the use of hatcheries to supplement natural production and increase population abundance for declining salmon populations (Flagg et al. 1999;Waples and Drake 2004;Naish et al. 2007;Paquet et al. 2011). While numerous salmonid supplementation programs have proven their value in providing a boost to declining natural populations (e.g., Fast et al. 2015;Janowitz-Koch et al. 2019), controversy remains over the best practices to maintain the ecological and genetic integrity of natural populations (Weber and Fausch 2003;Fraser 2008;Christie et al. 2012a).
Measuring fitness (i.e., offspring number) differences between fish born and reared in a hatchery setting (hatchery-origin) versus those in nature (natural-origin), along with the fitness of their surviving offspring, is a fundamental component of assessing the effectiveness and sustainability of supplemented populations. A wide range of studies spanning salmonid species, ranges, and timespans provide evidence that hatchery-origin fish demonstrate lower fitness than natural-origin fish (Kostow et al. 2003;McGinnity et al. 2003;McLean et al. 2003McLean et al. , 2004Araki et al. 2007a, b;2009;Williamson et al. 2010;Berntson et al. 2011;Thériault et al. 2011;Anderson et al. 2013;Milot et al. 2013;Evans et al. 2015;Sard et al. 2015;Neff et al. 2015;Ford et al. 2016;O'Sullivan et al. 2020) with the caveat that these effects could be partially mitigated by incorporating natural-origin fish into the hatchery broodstock (i.e., "integrated broodstock"; Hess et al. 2012;Ford et al. 2016;Janowitz-Koch et al. 2019). Although numerous studies have demonstrated declines in fitness for hatchery-origin fish, the causal mechanisms behind this decline have been difficult to elucidate and likely reflect various historical, ecological, and genetic factors (reviewed in Naish et al. 2007;Araki et al. 2008;Christie et al. 2014;Koch and Narum 2021). Despite declining fitness for hatchery-origin fish that can occur even after a single generation of hatchery rearing (Christie et al. 2012b), genome-wide genetic differences between hatchery-origin and natural-origin fish have remained undetected in many study systems, which may largely reflect continued gene flow between the hatchery and natural populations (Chittenden et al. 2010;Waters et al. 2015;Le Luyer et al. 2017;Gavery et al. 2018;López et al. 2019;Leitwein et al. 2021;Nilsson et al. 2021;Wellband et al. 2021). Rapid and heritable epigenetic changes in response to hatchery rearing could represent an explanation for fitness differences.
The burgeoning field of epigenetics has elucidated the relationship between environmental signals and transmittable phenotypic changes (reviewed in Weinhold 2006;Allis and Jenuwein 2016;Cavalli and Heard 2019). Broadly, the term epigenetics can be used to describe heritable nongenetic changes that may impact gene expression, but which do not alter the DNA sequence. Three of the most widely studied epigenetic mechanisms include DNA methylation, histone modification, and noncoding RNA. These epigenetic mechanisms can modulate gene expression through suppression of transcription, modification of chromatin structure, or post-transcriptional effects, and have the ability to control expression of transposable elements to create new genetic variation (reviewed in Slotkin and Martienssen 2007;Lisch 2009;Cavalli and Heard 2019). The transmission capability of epigenetic changes across multiple generations remains uncertain, but numerous studies throughout the taxonomic spectrum have provided evidence that epigenetic modifications incorporated into the germline can be meiotically stable and heritable (reviewed in Jablonka and Raz 2009;Skinner 2016;Bošković and Rando 2018;Anastasiadi et al. 2021).
While the epigenetic literature is heavily weighted toward mammals and plants, recent studies have detailed the contribution of epigenetic mechanisms modulating phenotypic responses for aquatic species in both aquaculture settings (reviewed in Moghadam et al. 2015;Gavery and Roberts 2017) and natural marine environments (reviewed in Eirin-Lopez and Putnam 2019). Aquatic species can provide unique opportunities largely unavailable in mammalian systems, such as the provision of a large number of gametes, as well as the ability to parse maternal from paternal epigenetic effects in oviparous fish, particularly in those species that lack parental care (reviewed in Best et al. 2018;Anastasiadi et al. 2021). Additionally, there are numerous well-characterized model fish species that are ideal for studying the relationship between epigenetics, development, environment, and phenotypic expression (Best et al. 2018).
Epigenetic modifications in fish can potentially arise as a direct result of environmental stressors, such as changes in temperature, density, predation stress, or even water quality parameters that can occur during key developmental stages and directly affect phenotypic plasticity, the emergence of lifehistory variables, and fitness (reviewed in Jonsson and Jonsson 2014). For example, threespine stickleback (Gasterosteus aculeatus) ecotypes exhibit differential methylation patterns, which overlap with genes related to osmoregulation and metabolism, and these patterns appear stable across generations, presumably facilitating adaptation to vastly different habitats (Heckwolf et al. 2020;Hu et al. 2021). Epigenetic responses to stressors that occurred in previous generations may therefore be heritable if erasure and reprogramming following fertilization or during gametogenesis do not occur, and this lack of reprogramming may occur more readily in teleost fish compared to mammals (Jiang et al. 2013;Manjrekar 2017;Ortega-Recalde and Hore 2019;Skvortsova et al. 2019). Due to this potential for heritability, there remain many unanswered questions as to what role environmentally induced epigenetic effects may play in acclimation and adaptation, particularly when environments change across generations (Heard and Martienssen 2014;Laland et al. 2014;Anastasiadi et al. 2021). On a macroscale, epigenetic modifications may increase the capacity to adapt to the rapid alteration of environments by climate change. This was recently demonstrated within a coral reef fish, Acanthochromis polyacanthus, which showed differential methylation patterns at gene regions related to aerobic scope, and seemingly improved ability to contend with warmer environments when exposed to similar, increased temperatures for several generations (Ryu et al. 2018). However, as we narrow the scope within the context of this review -for hatchery-reared salmonids that must contend with several environments (captive and natural) within a single lifetime, and which may produce offspring that never enter a captive environment, the benefit of inheriting parental epigenetic marks remains unclear.
Given the unique conditions that hatchery-origin fish experience in the captive environment and the clear phenotypic and behavioral differences that emerge between hatchery-origin and natural-origin fish, along with a lack of consistent genetic differentiation between the two, epigenetic differentiation is a likely mechanism shaping these differences (Fig. 1). A recent emergence of studies has examined the epigenetic differences between hatchery-origin and natural-origin fish as a potential mechanism contributing to fitness differences. We provide a synopsis on these key studies, primary themes that have emerged in the literature, and provide potential directions for using Fig. 1 Schematic representation of the potential transmission of epigenetic modifications in hatcheryorigin and natural-origin salmonids. In the first generation (F 0 ), individuals may be born in different environments (hatchery or nature) and may therefore experience different patterns of epigenetic modifications (methylation, in this example). If these modifications occur in the germ cells, they may be passed on to offspring (F 1 ), influencing eventual offspring phenotype despite being born in a common environment (nature) epigenetic tools to answer questions regarding salmonid supplementation practices moving forward.

Key findings across epigenetic studies of hatchery salmonids
Peer-reviewed studies were chosen based on a search query in Google Scholar that included combinations of the terms "salmonid," "epigenetics," "methylation," "gene expression," "supplementation," "hatchery," "fitness," and "reproductive success." We also checked reference lists in the included articles to search for any additional publications. Since 2010, there have been eight published studies that directly compared methylation differences between hatcheryorigin and natural-origin salmonids and two additional studies that compared gene expression patterns (Table 1). Of those studies that surveyed for genomewide genetic differences (i.e., nucleotide sequence variation), none identified significant differentiation by origin (Le Luyer et al. 2017;Gavery et al. 2018;Leitwein et al. 2021;Nilsson et al. 2021;Wellband et al. 2021). However, all of the reviewed studies, with the exception of Blouin et al. (2010), found substantial differences in methylation patterns between hatchery-origin and natural-origin fish ( Table 1). The Blouin et al. (2010) study, however, utilized a technique (msAFLP) which is not sensitive to fine-scale methylation differences (Blouin et al. 2010;Schrey et al. 2013;Le Luyer et al. 2017). Otherwise, three of the studies used whole genome bisulfite sequencing, which provides the greatest coverage and resolution of the methylome, whereas the remaining studies used various methods (i.e., RRBS, MBD-seq, MeDIP-seq) that are generally considered robust at identifying regions of differential DNA methylation across the genome (see for example reviews such as Nair et al. 2011;Kurdyukov and Bullock 2016;Singer 2019).
Not only were there similar results across studies, but there were also parallel methylation patterns induced by hatchery rearing across geographically distant rivers within the same study (Le Luyer et al. 2017;Leitwein et al. 2021). However, there were no consistent patterns in hypermethylation versus hypomethylation in hatchery-origin fish compared to natural-origin fish. Overall, three studies demonstrated greater levels of hypermethylation in hatchery-origin relative to natural-origin fish (Le Luyer et al. 2017;Rodriguez Barreto et al. 2019;Leitwein et al. 2021), while two studies found the opposite pattern of greater hypomethylation in hatchery-origin fish (Gavery et al. 2019;Wellband et al. 2021). Noted caveats in these studies include the possibility that the differences in hypermethylation versus hypomethylation across studies were affected by the type of methylation technique used, type of tissue, and/or time of sampling (life stage).
Genes associated with these differentially methylated regions demonstrated overlapping patterns across studies in enrichment for specific biological functions ( Table 1). Genes that overlapped results from multiple studies included those involved in embryonic development, growth, smoltification, stress, and immune response (Le Luyer et al. 2017;Gavery et al. 2018Gavery et al. , 2019Rodriguez Barreto et al. 2019;Leitwein et al. 2021Leitwein et al. , 2022Nilsson et al. 2021;Wellband et al. 2021). While methylation differences would be expected to lead to differential gene expression that may have fitness consequences for hatchery-reared fish (e.g., Christie et al. 2016), a single study that compared the patterns of gene expression and methylation between hatchery-origin and natural-origin fish did not find a strong relationship (Leitwein et al. 2022).
For oviparous fish such as salmonids, epigenetic changes must be incorporated into the germline before the release of gametes for transmission to the next generation (reviewed in Anastasiadi et al. 2021). Thus, measuring methylation in sperm or eggs can assess the potential for paternal and maternal epigenetic inheritance, respectively. The potential for paternal epigenetic inheritance was addressed in six studies by measuring methylation in sperm (Gavery et al. 2018(Gavery et al. , 2019Rodriguez Barreto et al. 2019;Leitwein et al. 2021;Nilsson et al. 2021;Wellband et al. 2021) and some of the specific patterns in methylation between hatchery-origin and natural-origin fish remained consistent across parents and their offspring (Rodriguez Barreto et al. 2019;Wellband et al. 2021). It is important to note, however, that no studies measured methylation in eggs and recent evidence in salmonids suggests that methylation is a potential mechanism explaining maternal effects (Venney et al. 2020) that is likely to influence multigenerational plasticity in different environments (Venney et al. 2021b). Table 1 Primary studies that have compared epigenetic profiles between hatchery-origin and natural-origin fish and two secondary studies (Christie et al. 2016;Leitwein et al. 2022) that compared gene expression patterns. Hatchery-origin and natural-origin refer to fish that were born in a hatchery setting or in the wild, respectively, while hatcheryreared and natural-reared refer to fish that were reared, but not born, in a hatchery setting or in a natural setting, respectively. Studies are presented in chronological and then alphabetical order

Causes, consequences, and future directions of epigenetics of hatchery-origin fish
Differences between methylation patterns of hatchery-origin and natural-origin fish were consistently detected across studies, yet the potential mechanisms explaining these differences have been largely unexplored. We propose general hypotheses for these differences and provide potential avenues for future research.  Jonsson 2014, 2019). In salmonids, methylation patterns can change in response to factors such as embryonic temperature (Burgerhout et al. 2017), embryonic oxygen levels (Kelly et al. 2020), and diet during juvenile stages of development (Morán et al. 2013;Marandel et al. 2016;Panserat et al. 2017). Furthermore, general environmental or rearing conditions can affect methylation patterns. For example, one recent study found increased methylation in Coho Salmon (Oncorhynchus kisutch) reared in a simulated stream environment compared to tank-reared fish (Christensen et al. 2021) and another study found that methylation patterns in Rainbow Trout (O. mykiss) can be impacted by the type of substrate in rearing tanks (Reiser et al. 2021). There is also evidence that the potential effects of hatchery rearing may be mutable across developmental stages or as conditions change.
For example, while Reiser et al. (2021) found an effect of tank substrate on methylation, these patterns did not persist when fish were sampled at later stages of development. Similarly, Gavery et al. (2019) found methylation differences between Steelhead Trout (O. mykiss) reared in traditional hatchery tanks versus in a simulated stream-rearing environment, but the patterns in methylation changed after a year in a common environment. However, Leitwein et al. (2021) demonstrated temporal persistence of hatchery-induced epigenetic marks into adulthood even after oceanic migration in Coho Salmon. The differences in findings between Gavery et al. (2019) and Leitwein et al. (2021) could reflect the resolution of the methylation techniques used, experiment design, or species differences. As a whole, these studies suggest that conditions experienced in the hatchery environment are likely to impact the epigenome, but these alterations can also be plastic and may not always persist within and across generations. While there have been a limited number of studies that have unveiled some of the factors that are likely shaping epigenetic differences between fish born and reared in hatcheries versus the natural environment, future studies using carefully controlled experimental designs will be a necessary next step toward uncovering the vast array of likely environmental drivers behind these epigenetic differences.

The relationship between epigenetic variation and gene expression
While methylation has generally been associated with inhibition of gene expression, the relationship between methylation, gene expression, and phenotypic response is dynamic and does not always display a predictable pattern ( In studies that have directly measured the relationship between gene expression and methylation between hatchery-origin and naturalorigin salmonids, the results have not been straightforward. Christie et al. (2016) found differential expression of hundreds of genes in Steelhead Trout offspring reared in a common environment from parents of different ancestry (hatchery-versus naturalorigin). Similarly, Leitwein et al. (2022) found thousands of genes that were differentially expressed in parallel across river systems of hatchery-origin versus natural-origin adult Coho Salmon. However, Leitwein et al. (2022) did not find a strong relationship between gene expression and methylation, with only a small proportion of genes demonstrating a correlation, suggesting other epigenetic modifications may be influencing the observed expression differences. These results were similar to another study that found little overlap between genes that were differentially methylated and those that were differentially expressed between Coho Salmon reared in simulated stream or in tank environments (Christensen et al. 2021). The results from these studies provide evidence that the relationship between the transcriptome and the epigenome is complex. Future studies should be aimed at exploring the relationship between epigenetic changes at key developmental stages and gene expression to determine whether epigenetic marks have longer-lasting effects on gene expression not captured in previous studies (e.g., Leitwein et al. 2022). Overall, careful study design that controls for tissue-specific differences, factors in different stages of development, provides biological replication, and explores other potential epigenetic mechanisms affecting gene expression will be important for further understanding of epigenomic and transcriptomic interactions.

Patterns in functional genomic regions associated with epigenetic responses to hatchery rearing
The field of epigenetics has transformed the genomics landscape, providing novel insight into the full picture of the functional genome and surpassing the utility of candidate gene approaches alone (reviewed in Feinberg 2010). While results across the salmonid studies reviewed here found substantial methylation changes in response to hatchery rearing, there were also very strong patterns of functional overlap between genes associated with these differentially methylated regions. One key theme that emerged across studies was differential methylation of genomic regions associated with stress and immune response (Le Luyer et al. 2017;Gavery et al. 2019;Rodriguez Barreto et al. 2019;Leitwein et al. 2021Leitwein et al. , 2022Nilsson et al. 2021;Wellband et al. 2021), including genes coding for corticotropin-releasing hormone receptors, an essential hormonal component of stress response (De Groef et al. 2006). There are numerous factors in the hatchery setting that could impact stress. For example, increased density in salmonid hatchery stocks can have an effect on various fitness-related traits including not only stress, but also immunity, growth, behavior, and post-release survival Rosengren et al. 2017). Further, early-life stressors such as temperature shock and lack of tank enrichment (both acute and chronic) have been shown to directly impact methylation and expression of genes that are associated with growth, development, and immunity (Burgerhout et al. 2017;Moghadam et al. 2017;Uren Webster et al. 2018). While it is uncertain why genes related to stress and immune response could be responding to hatchery rearing, it is possible that the higher density in the hatchery environment could increase likelihood of disease transmission, and potentially play a role in the first steps of genetic adaptation to captivity (Christie et al. 2016).
Other factors, such as temperature or diet could also represent catalysts for epigenetic change in response to hatchery rearing, as genes and pathways related to metabolism were found across studies (Christie et al. 2016;Gavery et al. 2019), including a study that factored in different dietary regimes that promoted slow or fast growth in hatchery-origin Steelhead Trout (Nilsson et al. 2021). Studies also found that insulin-like growth factor and epidermal growth factor genes were associated with differential methylation in hatchery-origin and natural-origin fish (Gavery et al. 2019;Wellband et al. 2021). Moreover, in a comparison of differentially methylated regions in hatchery-origin and naturalorigin Atlantic Salmon (Salmo salar) that overlapped phenotype-associated regions, Wellband et al. (2021) found that increased methylation at the gene encoding insulin-like growth factor I (IGF-1) was strongly associated with decreased weight and length. IGF-1 is a hormone that manages the effects of growth hormone (GH), playing a key role in directing bone and tissue growth across vertebrates (Ohlsson et al. 2009). In teleost fish, expression of GH and IGF-I is affected by numerous factors such as diet and feeding practices, temperature, photoperiod, chemicals, oxygen levels, stocking density, and salinity (reviewed in Triantaphyllopoulos et al. 2020). The GH/IGF-I axis is also important in osmoregulation and smoltification in fishes, including anadromous salmonids (reviewed in Sakamoto and McCormick 2006). Unlike naturalorigin fish that maximize their time in freshwater before smoltification, hatchery programs frequently manipulate spawning and rearing conditions to accelerate juvenile growth rates, which can ultimately affect fitness-related traits (Berejikian et al. 2012;Ford et al. 2012;Larsen et al. 2013;Blouin et al. 2021). Thus, the hatchery environment may directly impact methylation and expression of genes that are key modulators of growth and maturity, which can have a strong effect on fitness in salmonids.
Other genes and genetic modules that were associated with methylation in hatchery-origin versus natural-origin salmonids included those involved in neural development and cell differentiation. Specifically, enrichment of TATA-binding proteins, key transcriptional regulating genes (Pugh 2000), and numerous other transcription factors associated with embryonic development, were found across studies examined here (Le Luyer et al. 2017;Gavery et al. 2019;Rodriguez Barreto et al. 2019). It is not surprising that a broad array of genetic modules associated with early development, particularly neural development, were associated with changes in methylation in hatcheryorigin fish. Brain development has been associated with rearing environment in salmonids, but the effects may be plastic and depend on developmental stage (Kihslinger and Nevitt 2006;Campbell et al. 2015;Näslund et al. 2017;Reiser et al. 2021).
Overall, the clear overlap in patterns across studies suggests that the hatchery environment generates a conserved epigenetic signature of selection that is associated with captivity and domestication in salmonids. If these epigenomic modifications affect transcription and expression of genes that are associated with fitness-related traits, then epigenetics could present a promising avenue for reducing deleterious effects of captivity in salmonids. Future work should be directed at exploring patterns of these methylation signatures with gene expression and phenotypic changes, and determining whether these epigenetic signatures are conserved across generations.
The potential role of other epigenetic mechanisms While DNA methylation is the most widely studied epigenetic modification, there are other critically important forms of epigenetic modification that can independently or synergistically regulate the transcriptome (reviewed in Gibney and Nolan 2010;Portela and Esteller 2010). While the role of other epigenetic modifications besides methylation has been evaluated in salmonid species, these studies have not focused on the role of hatchery rearing. For example, evidence in Rainbow Trout suggests that both microRNAs and long noncoding RNAs can regulate epigenetic marks and affect gene expression (Wang et al. 2016;Kuc et al. 2017;Paneru et al. 2017). Other epigenetic mechanisms, such as histone modifications in Atlantic Salmon play a potential role in modulating immune responses during thermal stress (Boltana et al. 2018). The roles of mitochondrial epigenetics and the relationship between mitochondrial function and epigenetics of nuclear DNA have also largely been unexplored in aquatic organisms (Eirin-Lopez and Putnam 2019). While not all of these epigenetic processes are required to understand broad epigenetic differences between hatchery-origin and natural-origin salmonids, it is likely that the rapid response to hatchery rearing can involve alternative epigenetic mechanisms that have yet to be explored.
The interplay between environmental cues, epigenetic selection, and adaptive evolution Across eukaryotes, environmentally induced transcriptional changes are associated with epigenetic mechanisms that can have an impact on both phenotypic responses and ultimately, on adaptative evolution (reviewed in Feil and Fraga 2012;Skinner 2015;Anastasiadi et al. 2021).
Briefly, under situations of environmental challenges and change, epigenetic variation can promote a rapid evolutionary response in the absence of standing genetic variation (Schmitz et al. 2011;Klironomos et al. 2013). This epigenetic process could give rise to an adaptive phenotype before genetic selection and a subsequent adaptive genotype has emerged (Bossdorf et al. 2008;Klironomos et al. 2013). Subsequently, during a period of relaxed genetic selection, overall standing genetic variation can increase; thereby epigenetic selection can act as a stepping stone to genetic selection (Richards et al. 2010;Klironomos et al. 2013;Ashe et al. 2021). Transposable elements (TEs) are one mechanism through which the production of genetic and phenotypic variation can be regulated over short timescales and fish genomes show the highest diversity of TE superfamilies across vertebrates (Ferreira et al. 2011;Chalopin et al. 2015;Gao et al. 2016). The interplay between epigenetic mechanisms and TEs can modulate gene expression in response to environmental stress and increase phenotypic plasticity in real time (Rey et al. 2016). These plastic changes can then be incorporated into the genome (Fedoroff 2012;Rey et al. 2016;Almeida et al. 2022).
For aquatic organisms, under conditions of environmental shifts in freshwater ecosystems, epigenetics can promote rapid phenotypic responses, and if these epigenetic changes are heritable, selection and adaptive evolution could occur (see detailed reviews such as Munday 2014;Jeremias et al. 2018;Eirin-Lopez and Putnam 2019). A recent study found a strong relationship between high temperature, hypoxia, and methylation across targeted genes in post-smolt Atlantic Salmon that were associated with gene expression, which provides evidence for epigeneticmediated physiological acclimation to environmental change (Beemelmanns et al. 2021). A study in Chinook Salmon (O. tshawytscha) also showed that early rearing environment (hatchery tanks or seminatural stream channels) affected gene-specific patterns of methylation (Venney et al. 2021b). However, it is important to mention that not all studies found these strong associations. Another study in Chinook Salmon failed to identify a significant relationship between freshwater environmental variables in the natural environment and methylation across targeted genes, results which might reflect methodological limitations or an artifact of sampling during the embryo stage of development where fish could be shielded from environmental effects (Venney et al. 2021a). Overall, there are numerous environmental stimuli that could interact with the epigenome of hatchery-and natural-origin salmonids to illicit phenotypic responses. It is also possible that epigenetic effects in response to hatchery rearing could increase overall transmittable phenotypic diversity and thus widen the potential for adaptive responses across a range of environments in offspring (i.e., diversified bet hedging; Donelan et al. 2020;McGuigan et al. 2021).
Finally, it is important to discuss another scenario where adaptive mismatches could occur when heritable epigenetic responses in the parental environment are not adaptive for offspring in future environments (reviewed in Munday 2014; Eirin-Lopez and Putnam 2019; Anastasiadi et al. 2021;McGuigan et al. 2021). In the context of salmonid supplementation programs where fish could experience vastly different environmental conditions within a single lifetime (hatchery versus natural environment), and where parents, offspring, and even grandoffspring are also likely to encounter different rearing environments (hatchery parents producing offspring in the natural environment, for example), multigenerational epigenetic responses could potentially be maladaptive. A study in Brook Charr (Salvelinus fontinalis) showed that offspring methylation patterns were more affected by the rearing temperature of their parents than their own rearing temperature, suggesting that epigenetic inheritance could prepare offspring for future climatic shifts or could be maladaptive if the offspring environment is too different from the parental environment (Venney et al. 2022). Whether these epigenetic responses can actually persist across multiple generations, and whether they are ultimately adaptive across multigenerational environmental landscapes, is an important evolutionary question that remains to be answered in the context of salmonid supplementation.

Practical applications and summary of future directions
If epigenetics provides a bridge between the environment and the phenotype, then an intriguing area of research is how epigenetics can be used to inform breeding practices of salmonids and potentially reduce the gap between fitness of hatcheryorigin and natural-origin fish. One top-down approach to reducing these fitness differences is simulating the natural environment in the hatchery setting as much as possible. Factors such as physical enrichment or density, to name a few, can potentially mitigate for fitness reductions in hatchery-origin fish (reviewed in Johnsson et al. 2014) and should be considered as potential drivers of epigenetic differentiation. This approach does not require a deep understanding of the entirety of factors within a hatchery setting that could be causing epigenomic changes; rather, it assumes that the hatchery environment is different from the natural environment and thus changes are made to reduce those differences whenever feasible.
A more bottom-up approach would include targeted selection on epigenetic markers that are linked to fitness-enhancing traits. Broodstock selection for epigenetic markers that are associated with acclimatization and adaptation, for example, could be an extremely useful tool to increase production of fish that would be better adapted to the natural environment (Eirin-Lopez and Putnam 2019). There are a myriad of other specific factors that could be targeted through epigenetic selection, such as disease resistance, growth, metabolism, and behavior that could be beneficial for influencing production practices and mitigating for fitness declines (Moghadam et al. 2015;Gavery and Roberts 2017). While a targeted epigenetic selection approach could be extremely valuable to enhancing fitness of salmonid broodstocks overall, it requires an in-depth understanding of the epigenetic association with fitness-related traits and the stability of epigenetic marks.
The relationship between the epigenome and expression of phenotypic traits is further complicated by the interaction between the genome and the epigenome. An obvious place where these two ideas could collide is in the type of broodstock production practices that are utilized in hatcheries. Salmonid hatcheries generally rely on using an integrated or a segregated model for broodstock collection. The segregated broodstock model incorporates only hatchery-origin fish into broodstock, thereby creating a hatchery population that is genetically distinct from the natural population, whereas the integrated broodstock model incorporates fish born in nature into the hatchery broodstock with the goal of increasing gene flow of natural-origin alleles and minimizing domestication (Mobrand et al. 2005;Paquet et al. 2011). Previous work has demonstrated clear differences from these two management approaches in the amount of genetic divergence from natural populations (Waters et al. 2015). If epigenomic variation interacts with genetic variation to predict phenotypes (i.e., genotype x epigenotype; Richards 2006;Rougeux et al. 2019;Anastasiadi et al. 2021), then it is possible that these two distinct broodstock models would have different expectations for phenotypic changes in response to hatchery rearing. This area of research is yet to be explored, but could provide baseline expectations for how broodstock collection practices may interact with the epigenome to shape phenotypic expression in salmonids.
Finally, epigenetic tools could be used to target very specific issues in salmonid supplementation. For example, disproportionally high levels of precocial male maturation have been directly linked to hatchery production (Larsen et al. 2004(Larsen et al. , 2021Harstad et al. 2014), and compared to larger anadromous males, precocial males typically have lower fitness (Berejikian et al. 2010;Ford et al. 2012;Schroder et al. 2012). However, the specific mechanisms driving precocial maturation have not been straightforward. One study compared methylation levels between Atlantic Salmon mature male parr and immature male parr and found distinct methylation patterns, particularly in the gonads (Morán and Pérez-Figueroa 2011). Therefore, the relationship between hatchery rearing and early male maturation could be further elucidated by using epigenetic tools, such as high-resolution methylation techniques, or exploration of other epigenetic mechanisms.
In summary, while the results from studies comparing epigenetic differences between hatcheryorigin and natural-origin salmonids are just beginning to emerge, there are numerous gaps that could help to further elucidate epigenetic contributions to fitness differences and for improving hatchery management practices (Fig. 2). First, it is important to understand the potential range of factors that could affect epigenetic mechanisms, particularly during early stages of development, as a result of hatchery rearing. Second, there are multiple types of epigenetic changes that can potentially occur in response to hatchery rearing besides or in addition to methylation. Further, results across studies suggest that differentially methylated loci or regions between hatchery-origin and natural-origin fish overlap functionally important genes across the genome. Thus, further exploration of the effect of epigenetic changes on the molecular phenotype is crucial. Third, epigenetic differences may not be evolutionary meaningful without an effect on phenotypic variation. Thus, the relationship between epigenetics and phenotypic manifestation, and the impact on fitness, is vital to determining the long-term significance of epigenetic differences between hatcheryorigin and natural-origin fish. Fourth, further exploration of the transmission capability of epigenetic marks is needed, including multigenerational inheritance, evolutionary implications of potential parent-offspring mismatches, and timeframes for resetting epigenetic marks. Overall, there are countless ways in which hatchery practices could be shaped and potentially enhanced by the field of epigenetics. As we look to the future, it is clear that epigenetics can be a powerful tool to inform management practices and will likely be a key step toward improving overall fitness of hatchery-origin salmonids.
Acknowledgements The authors would like to acknowledge the Bonneville Power Administration for funding.
Funding This work was supported by the Bonneville Power Administration.
Data availability Data sharing is not applicable to this article as novel datasets were not generated or analyzed during the current study.

Declarations
Research involving animals This review used only published literature and did not include any research on living or preserved animals requiring ethical clearance.

Competing interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.

Fig. 2
Conceptual figure highlighting the key areas of future research to explore epigenetic responses to hatchery rearing. Key questions are separated by the number of generations (F 0 and onward) that the question is most applicable to, and includes potential approaches to address those questions